What is the difference in between the inception v3 and Convolutional neural network?
$\begingroup$
$\endgroup$
1
-
1$\begingroup$ The Inception architecture is a convolutional model. It just puts the convolutions together in a more complicated (perhaps, sophisticated) manner, which allows the model to be a little more robust to the variance in size of the objects within images. $\endgroup$– n1k31t4Commented Aug 2, 2018 at 22:01
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
1
The Inception models are types on Convolutional Neural Networks designed by google mainly for image classification. Each new version (v1, v2, v3, etc.) marks improvements they make upon the previous architecture.
The main difference between the Inception models and regular CNNs are the inception blocks. These involve convolving the same input tensor with multiple filters and concatenating their results. Such a block is depicted in the image below.
On the contrast, regular CNNs performs a single convolution operation on each tensor.
Inception-v3 is Deep Neural Network architecture that uses inception blocks like the one I described above. It's architecture is illustrated in the figure below.
The parts where the layers "branch off" and then are merged together again are the inception blocks described previously.
You can read more about Inception-v3 here.
-
1$\begingroup$ The various filter sizes help the model to generalise for objects of different sizes, in that it offers a range of focus levels. Also worth noting is that the 1x1 convolutions are included in the 3x3 and 5x5 branches as a means to reduce the computational cost. they shrink the sizes of the tensors by reducing the third dimension in a 3d tensor. $\endgroup$– n1k31t4Commented Aug 2, 2018 at 22:04